Use menuconfig objects
cover Use menuconfig objects
Quoted 2 lines · Jan Engelhardt
the following patch series turns some menus into menuconfigs, so they can be disabled whilst "walking" thorugh the parent menu
Great, but instead of making it a simple on/off, make it tri-state that would default select all child-options appropriately. (see HW_RANDOM)
Thanks!
--
Al
Quoted 6 lines · Al Boldi
Quoted 2 lines · Jan Engelhardt
the following patch series turns some menus into menuconfigs, so they can be disabled whilst "walking" thorugh the parent menu
Great, but instead of making it a simple on/off, make it tri-state that would default select all child-options appropriately. (see
HW_RANDOM)
I do not remember having touched the Random Number Generators or Character Devices anywhere in this series.
Don't worry, I have paid attention (or at least I hope so ;-) to 'm'-able options, for example the SCSI, IEEE1394, or I2C stacks.
But, for example you will see
[*] Old CD-ROM drivers (not SCSI, not IDE)
which is because this entry in itself does not generate any object file.
Generally, I just made the 'menuconfig' entry have the same state (bool/tristate) as the original entry. If I have overseen that somewhere, please let me know!
Thank you,
Jan
Quoted 14 lines · Jan Engelhardt
Quoted 6 lines · Al Boldi
Quoted 2 lines · Jan Engelhardt
the following patch series turns some menus into menuconfigs, so they can be disabled whilst "walking" thorugh the parent menu
Great, but instead of making it a simple on/off, make it tri-state that would default select all child-options appropriately. (see
HW_RANDOM)I do not remember having touched the Random Number Generators or Character Devices anywhere in this series.
Don't worry, I have paid attention (or at least I hope so ;-) to 'm'-able options, for example the SCSI, IEEE1394, or I2C stacks.
Quoted 8 lines
But, for example you will see
[*] Old CD-ROM drivers (not SCSI, not IDE)which is because this entry in itself does not generate any object file.
Generally, I just made the 'menuconfig' entry have the same state (bool/tristate) as the original entry. If I have overseen that somewhere, please let me know!
Correct, no oversight here; but it may be more meaningful, if you could default select child-options based on the parent-state, like HW_RANDOM.
Also, I don't think it's necessary to touch any of the "depends on"; keep them as is, as they don't hurt staying that way, and may actually be necessary under certain circumstances. (see EMBEDDED)
Thanks!
--
Al
Quoted 4 lines · Al Boldi
Jan Engelhardt wrote:
Correct, no oversight here; but it may be more meaningful, if you could default select child-options based on the parent-state, like
HW_RANDOM.
A symbol will already have the 'magic' you describe if its creator wrote it that way. Changing to 'menuconfig' tokens does not change that I think. Consider:
menu "foobar"
config FOOBAR
bool "foobar"
default y
config HW_RANDOM
depends on FOOBAR
default y
endmenu
This example should do things as you describe -- if FOOBAR is changed from n to y, HW_RANDOM should automatically do the same since it is a new option then.
menuconfig FOOBAR
bool "foobar"
default y
if FOOBAR
config HW_RANDOM
default y
endif
inhibits the same behavior AFAICS.
Quoted 3 lines
Also, I don't think it's necessary to touch any of the "depends on"; keep them as is, as they don't hurt staying that way, and may actually be necessary under certain circumstances. (see EMBEDDED)
Simplifying the depends lines is a good thing IMO.
Jan
Quoted 6 lines · Jan Engelhardt
Quoted 3 lines · Al Boldi
Also, I don't think it's necessary to touch any of the "depends on"; keep them as is, as they don't hurt staying that way, and may actually be necessary under certain circumstances. (see EMBEDDED)
Simplifying the depends lines is a good thing IMO.
If explicit dependency statements are replaced by if--endif blocks, does this count as simplification?
Signature · 3 lines
Stefan Richter -=====-=-=== -=-- -=-== http://arcgraph.de/sr/
Quoted 10 lines · Stefan Richter
Quoted 6 lines · Jan Engelhardt
Quoted 3 lines · Al Boldi
Also, I don't think it's necessary to touch any of the "depends on"; keep them as is, as they don't hurt staying that way, and may actually be necessary under certain circumstances. (see EMBEDDED)
Simplifying the depends lines is a good thing IMO.
If explicit dependency statements are replaced by if--endif blocks, does this count as simplification?
According to Documentation/kbuild/, if FOO .. endif is interpreted as if each option inside the if block had 'depends on FOO'.
If _every_ option under a menu (and that's another reason why I did not modify _all_ Kconfig menus) has a 'depends on this or that', it can be substituted by an if block, and yes, I take that as a simplification since the whole LoC count goes down. What would perhaps be cool is some indent in Kconfig to see what an if-endif block spans, but currently this is not much needed, since two nested ifs are quite the max.
Jan
Quoted 10 lines · Stefan Richter
Quoted 6 lines · Jan Engelhardt
Quoted 3 lines · Al Boldi
Also, I don't think it's necessary to touch any of the "depends on"; keep them as is, as they don't hurt staying that way, and may actually be necessary under certain circumstances. (see EMBEDDED)
Simplifying the depends lines is a good thing IMO.
If explicit dependency statements are replaced by if--endif blocks, does this count as simplification?
Applied with some common sense - yes.
if --endif blocks has been used in several places to make dependencies consistent. When cleaning up the net part I recall a few places where the explicit dependency was missing causing 1) dependencies to be wrong and 2) caused indention in menu- config to be wrong.
But yet agin do not go and replace all dependencies with if--endif blocks. They shall be used with some good taste.
Sam
Quoted 19 lines · Jan Engelhardt
Quoted 4 lines · Al Boldi
Jan Engelhardt wrote:
Correct, no oversight here; but it may be more meaningful, if you could default select child-options based on the parent-state, like
HW_RANDOM.A symbol will already have the 'magic' you describe if its creator wrote it that way. Changing to 'menuconfig' tokens does not change that I think. Consider:
menu "foobar"
config FOOBAR
bool "foobar"
default yconfig
HW_RANDOM
depends on FOOBAR
default y
Quoted 5 lines
endmenu
This example should do things as you describe -- if FOOBAR is changed from n to y,
HW_RANDOMshould automatically do the same since it is a new option then.
I don't think so. But this might:
menu "foobar"
config FOOBAR
bool "foobar"
default y
config HW_RANDOM
depends on FOOBAR
default FOOBAR
endmenu
Quoted 5 lines
menuconfig FOOBAR
bool "foobar"
default y
Quoted 8 lines
if FOOBAR
config
HW_RANDOM
default yendif
inhibits the same behavior AFAICS.
I think you mean "exhibits"?
Quoted 6 lines
Quoted 4 lines
Also, I don't think it's necessary to touch any of the "depends on"; keep them as is, as they don't hurt staying that way, and may actually be necessary under certain circumstances. (see EMBEDDED)
Simplifying the depends lines is a good thing IMO.
You are replacing the depends with an if block, which seems to add complexity, as we now have two constructs, if and depends.
IMHO, leave the depends as is, for simplicity and correctness and maintainability.
So if we have something like this:
menu "foobar"
config FOOBAR
bool "foobar"
default y
config FOOBAR_SUB1
tristate "FOOBAR SUB1"
depends on FOOBAR
default m
config FOOBAR_SUB2
bool "FOOBAR SUB2"
depends on FOOBAR
default y
endmenu
Change it to something like this:
menuconfig FOOBAR
tristate "foobar"
default y
config FOOBAR_SUB1
tristate "FOOBAR SUB1"
depends on FOOBAR
default FOOBAR
config FOOBAR_SUB2
bool "FOOBAR SUB2"
depends on FOOBAR
default FOOBAR
without any if blocks, and maybe even in its own Kconfig.
Thanks!
--
Al